And looking at the flash, and the effects (or not) that it had on the solid state imagers, I think we can rule out a nuclear intercept. My reasoning is that a 10kT nuclear detonation @ 20km would toast any unprotected CCD or CMOS image element looking at it.

I assume there would have also been a detectable gamma burst that Western satellites would've picked up if they're trained on the region.

There was an interesting (if short) conversation on another list about how the US was able to spot small scale nuclear events (even things like a criticality event in a poorly shielded reactor) The speculation revolved around some unidentified sensors that were on the GPS satellite at the San Diego Air & Space museum, since removed (the sensors not the model)) and the way in which GPS satellites could tell you where things were relative to them, just as easily as your car can figure out where it is relative to the satellite.

There are a number of monitoring systems spelled out in the various disarmament treaties, and the infrasound sensors were part of the test ban treaty, so I don't doubt that if there had been a nuclear component, someone would know about it :-)

So far the Arms Control Wonk blog hasn't said anything about it.

I wonder if it's possible that it was an automated response?

Some combination of mass and speed detected by radar (or satellite?) triggering an S2A response.

Not a chance. Taking into account:

   - the size of Russia

   - The range of surface to air missiles

   - the chance of missing an object that does not give off
     a heat signature until it is about to hit.

Any kind of suggestion that we have the tech or the means (financially) to do this for a large chunk of the planet is nonsense, and for a smaller area the odds of hitting something are probably so bad that it isn't worth it.

Yeah but this isn't a random location. Chelyabinsk being home to Russia's nuclear research, and relatively close to Baikonur.